Join GitHub today
GitHub is home to over 31 million developers working together to host and review code, manage projects, and build software together.Sign up
Axis autotype update #3070
Fixes #3039 - treat
Fixes #2473, fixes #1413, by counting only distinct values while determining date and category autotype, as discussed in #2473 (comment). This way we don't need to include any explicit "missing" values; data with up to 2 non-numeric values for every numeric value will still be interpreted as numbers, but even a single non-numeric string (including the previous special values
Note that as part of ^^ I also converted the date determination to counting distinct values. Nobody had complained about this part, but if I didn't do that there could be some strange cases where you have only date strings and numbers but our result is
There's probably a way to improve performance (and make it clearer that we can't generate such strange results) by combining
referenced this pull request
Oct 3, 2018
Hmm. I would be nice to check with a large splom (e.g. https://codepen.io/etpinard/pen/wjmqmO). You're right, looping over <1000 points shouldn't be too slow, but looping over <1000 pts once per axis ~100 times could have an impact. Related: #2549
Note: I'll check this myself off my latest splom commits.
The results are in: these commits here don't make much of a difference in axis-autotype perf.
But I never realized how much time large splom traces spend in